Conceptos Básicos
The Joint Weighted Average (JWA) operator systematically integrates a priori beliefs about the quality of information sources and the quality of the evidence arising from those sources to enable more nuanced and robust information aggregation.
Resumen
The paper introduces the Joint Weighted Average (JWA) operator, a novel approach for aggregating information that jointly considers the quality of the information sources and the quality of the evidence produced by those sources.
The authors first review existing aggregation operators, such as the Linear Weighted Average (LWA) and the Order Weighted Average (OWA), which focus on source quality and evidence quality respectively. They identify limitations in prior attempts to combine these two approaches, such as the Weighted OWA (WOWA) and Hybrid Weighted Average (HWA), which do not fully integrate the two strategies.
The authors then leverage the mathematical framework of compositional geometry to develop the JWA operator. This allows them to systematically blend the linear weights reflecting source quality with the order weights reflecting evidence quality. The resulting joint weights represent the combined worth of both the sources and the evidence.
Through examples and simulations, the authors demonstrate how the JWA outperforms prior aggregation operators by capitalizing on the strengths of both the source-focused and evidence-focused strategies, while remaining robust to challenging data contexts that degrade the performance of the individual strategies.
The authors conclude by highlighting the potential of the JWA to bridge the gap between human and artificial reasoning, as well as its applications in areas like healthcare, machine learning, and decision support systems.
Estadísticas
The evidence from 10 sources has varying validity, with some sources being more predictive of the criterion variable than others.
The evidence from some sources is also subject to a random positive bias in 50% of the trials.
Citas
"The JWA jointly focuses on source-and-evidence specific worth by directly combining the linear and ordered weights using compositional geometry [1,14,13]."
"This approach reveals how previous work attempting to combine LWA and OWA have suggested that a combined operator must adhere to a set of mathematical properties that precludes this joint focus."